A system and method for rerouting circuits on a SONET, SDH or similar ring without disrupting service. In one embodiment, the system includes: (1) a route design subsystem adapted to identify, for the circuits, alternative routes that are subject to an alignment constraint and increase an objective value associated with the ring and (2) a path migration subsystem, associated with the route design subsystem, adapted to determine a hitless migration sequence that allows the alternative routes to be employed.
|
11. A method of rerouting circuits on a ring, comprising:
defining a sequence of time slots, each time slot corresponding to a circuit capacity unit;
identifying, for a circuit on said ring, an alternative route subject to an alignment constraint and assigned a number of contiguous time slots in said sequence, wherein said number of contiguous time slots corresponds to multiple circuit capacity units required by said circuit, and said assigned time slots start at one of given pre-selected specific time slots in said sequence pursuant to said alignment constraint; and
determining a hitless migration sequence that allows said alternative route to be employed;
wherein said method reroutes multiple circuits at any given time.
21. A system for rerouting circuits on a ring, comprising:
means for defining a sequence of time slots, each time slot corresponding to a circuit capacity unit;
means for identifying, for a circuit on said ring, an alternative route subject to an alignment constraint and assigned a number of contiguous time slots in said sequence, wherein said number of contiguous time slots corresponds to multiple circuit capacity units required by said circuit, and said assigned time slots start at one of given pre-selected specific time slots in said sequence pursuant to said alignment constraint; and
means for determining a hitless migration sequence that allows said alternative route to be employed;
wherein said system is configured to reroute multiple circuits at any given time.
1. A system for rerouting circuits on a ring, comprising:
a route design subsystem for defining a sequence of time slots, each time slot corresponding to a circuit capacity unit, an alternative route for a circuit on said ring being identified subject to an alignment constraint and being assigned a number of contiguous time slots in said sequence, wherein said number of contiguous time slots corresponds to multiple circuit capacity units required by said circuit, and said assigned time slots start at one of given pre-selected specific time slots in said sequence pursuant to said alignment constraint; and
a path migration subsystem, associated with said route design subsystem, that determines a hitless migration sequence that allows said alternative route to be employed;
wherein said route design and path migration subsystems are configured to reroute multiple circuits at any given time.
2. The system as recited in
3. The system as recited in
4. The system as recited in
5. The system as recited in
where Pi is a packing set for granularity i and g(ck) and l(ck) are granularity and length, respectively of circuit ck in Pi, and f is a heuristic function.
6. The system as recited in
rerouting said ones to a different time slot,
reversing a direction of said ones, and
rerouting by both rerouting said ones to said different time slot and reversing said direction.
7. The system as recited in
8. The system as recited in
a Unidirectional path-Switched ring (UPSR) in which said circuits are unprotected, and
a Bidirectional Line-Switched ring (BLSR).
9. The system as recited in
1,
3,
12,
48, and
192.
10. The system as recited in
Synchronous Optical NETwork (SONET), and
Synchronous Digital Hierarchy (SDH).
12. The method as recited in
13. The method as recited in
14. The method as recited in
15. The method as recited in
to determine an objective value associated with said ring, where Pi is a packing set for granularity i and g(ck) and l(ck) are granularity and length, respectively of circuit ck in Pi, and f is a heuristic function.
16. The method as recited in
rerouting said ones to a different time slot,
reversing a direction of said ones, and
rerouting by both rerouting said ones to said different time slot and reversing said direction.
17. The method as recited in
18. The method as recited in
a Unidirectional path-Switched ring (UPSR) in which said circuits are unprotected, and
a Bidirectional Line-Switched ring (BLSR).
19. The method as recited in
1,
3,
12,
48, and
192.
20. The method as recited in
Synchronous Optical NETwork (SONET), and
Synchronous Digital Hierarchy (SDH).
22. The system as recited in
23. The system as recited in
24. The system as recited in
|
The present invention is directed, in general, to multiplexed communication links and, more specifically, to a system and method for rerouting circuits on Synchronous Optical NETwork (SONET) or similar rings without disrupting service.
The last few years have seen a significant growth in of Synchronous Optical NETwork (SONET) and Synchronous Digital Hierarchy (SDH) deployments by telecommunications service providers. (Since the present invention applies to SONET, SDH and similar network protocols equally well, only SONET networks will be explicitly discussed hereinafter.)
While initially intended for voice traffic, SONET networks have come to provide the underlying transport for the growing Internet data traffic. Unfortunately, unlike voice, data traffic has more variance and therefore creates greater “churn” (a building up and tearing down of links) in the network. As the data traffic continues to grow and the capital budgets of service providers fail to keep up (or even diminish), providers are increasingly seeking network engineering tools that enable them to extract higher utilization from their existing infrastructure.
Referring initially to
Turning now to
As beneficial as it has been demonstrated to be, network engineering suffers a critical constraint. Since it is performed on operational networks that carry live traffic, it should be hitless, i.e., cause no service disruption. Thus, in addition to optimizing the layout of circuits, it is equally important to determine a hitless rerouting sequence to migrate the ring from the original to this new layout. Otherwise, rerouting is of little practical use.
In traditional SONET rings consisting of add-drop elements, rerouting traffic was often a cumbersome task and in many cases, impossible to achieve without disruption. However, newer SONET network elements are increasingly supporting traditional add-drop capability with a more mesh-like cross-connect functionality. Thus, these new element can support a “bridge-and-roll” functionality (see, M.3100, “Definition of the Management Interface for a Bridge-and-Roll Cross-Connect Feature,” Amendment 4, ITU-T, August 2001, incorporated herein by reference in its entirety). This enables circuits to be first bridged on to the new route and then rolled over seamlessly with no service interruption, akin to make-before-break in MPLS (see, e.g., Awduche, et al., “RSVP-TE: Extension to RSVP for LSP Tunnels,” IETF RFC 3209, 2001, incorporated herein by reference in its entirety).
Three ways exist to reroute circuits on a SONET ring: (1) moving a circuit to a different time slot (e.g., circuit C), (2) reversing the direction of the route of the circuit (e.g., from a clockwise to a counter-clockwise direction (e.g., circuit D) and (3) doing both (e.g., circuit B). Thus, rerouting circuits C, D and B (in that order) using bridge-and-roll ensures a migration with no perceived service hit.
The requirement to provide a disruption-free transition sequence fundamentally differentiates network engineering from network design. Moreover, it precludes periodically redesigning the circuit routes to achieve the same goal. Network engineering, thus, requires addressing two critical, yet distinct problems: “route design” and “path migration,” both of which are theoretically “hard” problems to solve. Scaling network engineering to larger rings (e.g., STS-48/192) with more nodes proves an extremely difficult challenge.
Extensive research has been done in the area of SONET ring design. For example, Cosares, et al., “An Optimization Problem Related to Balancing Loads on SONET Rings,” Telecommunication Systems, vol. 3, 1994, incorporated herein by reference in its entirety; Myung, et al., “Optimal Load Balancing On SONET Bidirectional Rings,” Operations Research, vol. 45, January 1997 incorporated herein by reference in its entirety; and Schrijver, et al., “The Ring Loading Problem,” SIAM Journal of Discrete Math, vol. 11, no. 1, February 1998, incorporated herein by reference in its entirety, have studied the SONET ring loading portion of the problem and developed efficient heuristic algorithms. The portion of the problem that involves assigning time slots given a routing, is known to be NP-hard (see, e.g., Garey, et al., Computers and Intractability—A Guide to the Theory of NP-Completeness, Freeman, Calif., USA, 1979, incorporated herein by reference in its entirety) and follows from the circular arc graph coloring problem. A comprehensive treatment of all the aspects of SONET ring design is available in Carpenter, et al., “Demand Routing and Slotting on Ring Networks,” DIMACS technical Report, Tech. Rep., January 1997 incorporated herein by reference in its entirety; and Cosares, et al., “Network Planning with SONET Toolkit,” Bellcore Exchange, September 1992, incorporated herein by reference in its entirety.
Unfortunately, these contributions do not account for the alignment property, which will be described in detail below. Unlike ring design, very little prior work has been undertaken on the hitless path migration problem. Shepherd, et al. (“Hardness of Path Exchange Problem,” Personal Communication, incorporated herein by reference in its entirety) proved that path migration is NP-hard in the general case of a mesh network.
Accordingly, what is needed in the art are effective systems and methods for achieving hitless network engineering with respect to SONET (and similar) rings.
To address the above-discussed deficiencies of the prior art, the present invention provides a system for rerouting circuits on a ring. In one embodiment, the system includes: (1) a route design subsystem adapted to identify, for the circuits, alternative routes that are subject to an alignment constraint and increase an objective value associated with the ring and (2) a path migration subsystem, associated with the route design subsystem, adapted to determine a hitless migration sequence that allows the alternative routes to be employed.
In another aspect, the present invention provides a method of rerouting circuits on a ring. The method includes: (1) identifying, for the circuits, alternative routes that are subject to an alignment constraint and increase an objective value associated with the ring and (2) determining a hitless migration sequence that allows the alternative routes to be employed.
The present invention therefor introduces an efficient engineering solution for SONET (and similar) rings that takes the alignment property (to be described below) into account. Instead of infrequently engineering the network as done today due to high overhead, the present invention enables providers to perform network engineering proactively. Proactive network engineering maintains high levels of efficiency all through a network, saving capital over time.
The foregoing has outlined, rather broadly, preferred and alternative features of the present invention so that those skilled in the art may better understand the detailed description of the invention that follows. Additional features of the invention will be described hereinafter that form the subject of the claims of the invention. Those skilled in the art should appreciate that they can readily use the disclosed conception and specific embodiment as a basis for designing or modifying other structures for carrying out the same purposes of the present invention. Those skilled in the art should also realize that such equivalent constructions do not depart from the spirit and scope of the invention in its broadest form.
For a more complete understanding of the present invention, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:
It has been found that network engineering with respect to SONET (and similar) rings involves the solution of a problem referred to herein as the Ring Rearrangement Problem (RRP). Before formally describing RRP, five assumptions derived from SONET/SDH constraints that pertain to RRP will be highlighted. For complete details on SONET/SDH, the reader is referred to Goralski, supra. The five assumptions are as follows:
(1) Demand Granularity: Circuits are assumed to follow SONET STS hierarchy, namely that circuits are one of STS-<1,3c,12c,48c,192c> with each STS-nc circuit carrying n times the capacity of a STS-1 circuit.
(2) Alignment property: SONET also imposes an alignment property, namely that a STS-nc circuit can only use slots s to (s+n−1) such that s≡1(mod n). In other words, a STS-3c circuit can be provisioned starting only at slots 1, 4, 7, 10, etc., and a STS-12c circuit can be provisioned only at slots 1, 13, 25, etc. Surprisingly, in spite of its criticality, this alignment property has rarely been considered in the body of prior art directed to loading and sizing SONET systems (e.g., Cosares, et al.; Myung, et al.; and Schrijver, et al., supra).
(3) BLSRs: The engineering of BLSRs will be the focus herein. Unlike Unidirectional Path-Switched Rings (UPSRs), BLSRs enable higher utilization by routing each circuit only along one side, and allowing protection bandwidth to be shared among circuits. However, those skilled in the pertinent art will be able readily to apply the principles of the present inventions to unprotected circuits on UPSRs.
(4) Time Slot Assignment (TSA): In a TSA ring system, a circuit is required to follow the same slots on all the links along its path from the source to the destination node. A TSA system has a less complex logic for protection switching and is enforced in the well-known Bellcore standard GR-1230-CORE (“SONET Bidirectional Line Switched Ring Equipment Generic Criteria,” Bellcore, November 1995, incorporated herein by reference in its entirety).
(5) Bridge-and-Roll: Support for Bridge-and-Roll functionality to provide hitless migration will be assumed.
Having set forth the five underlying assumptions, RRP is now defined as follows:
Solving RRP involves addressing two problems, namely route design and path migration problems, which correspond to each of the two requirements. The first is a variant of the well-known demand routing and slotting problem (DRSP) (see, Carpenter, et al., supra). The ring sizing and the ring loading are specific examples of the DRSP and these are proven to be NP-hard.
Good approximations to these problems do exist (e.g., Cosares, et al.; Myung, et al.; and Schrijver, et al., supra). Unfortunately, these solutions do not suffice for RRP. They do not account for the alignment property, which complicates the slotting. Even if one were able to account efficiently for alignment, the second requirement of a hitless migration sequence still remains. Unfortunately, such a sequence may not exist at all and in general, it has also been shown to be NP-hard (see, Shepherd, et al.). Thus, a two step approach to solving RRP is of only limited value in practice where a hitless migration is key.
A novel algorithm will now be introduced and followed by a description of the objective function that the algorithm takes as an input. The pseudocode for the algorithm is presented in FIG. 2. The algorithm is iterative and attempts to reroute circuits one at a time.
For every circuit in the network, the algorithm determines all possible ways to reroute it on the existing network. Recall that as described above, three ways exist for doing this. For each reroute, the algorithm determines the change in the objective function. The algorithm does so for all the circuits, and the one that provides the maximum increase in the objective value, is chosen to be rerouted. This process repeats until no reroute improves the objective value, at which point the algorithm terminates.
Since the algorithm makes a greedy choice at every step, it is possible that it settles on a local maximum for the objective function. Finding the global maximum is the desired goal. However, as pointed out above, ensuring global optimality is a theoretically hard (NP-hard) problem and therefore, unlikely to be achievable in practice. Besides, from the service provider's perspective, any improvement is desirable particularly if it creates no service disruption. In fact, experiments described below demonstrate that the algorithm is indeed effective in practice.
From a service provider's perspective, the goal of network engineering is to increases the revenues from the existing infrastructure. This suggests that the layout of the free time slots on the ring mirror the future demand pattern. Thus, an effective objective function is one that accurately valuates the “worth” of the layout of free slots. However, determining a “good” layout is not always obvious.
Turning again to the two networks represented in
The concept of a “packing set” is defined to address the first problem. A packing set of granularity g is the minimum number of new circuits of bandwidth g that would pack the ring to the fullest extent possible.
Packing sets provide a fair metric, since they use the minimum number of new circuits to enumerate the ability to satisfy future requests. Determining the packing set is a straightforward task for those skilled in the pertinent art; thus, the details are omitted here.
Service providers typically have some broad knowledge of future demands. In the absence of this domain knowledge, a heuristic function f is used to balance among circuits of different granularities and different span lengths. Various options for f will be explored below.
Given this background, the illustrated embodiment of the algorithm uses the following objective function Z to evaluate the worth of a network:
where Pi is the packing set for granularity i, and g(ck) and l(ck) are the granularity and length respectively of circuit ck in Pi. Thus, given a network, the packing set is first determined for different granularities that the network can support. Then, the function is evaluated to determine its worth. This provides the basis for comparison among different ring layouts. The objective function is orthogonal to the algorithm itself. Those skilled in the pertinent art understand that any alternative function may be provided to the algorithm without departing from the broad scope of the present invention.
Turning now to
The system 400 further includes a path migration subsystem 420. The path migration subsystem 420 is associated with the route design subsystem 410 and is adapted to determine a hitless migration sequence that allows the alternative routes to be employed. The route design subsystem 410 and the path migration subsystem 420 may be embodied in a sequence of computer-executable instructions that may function in general accordance with the novel algorithm set forth above.
Turning now to
The method 500 then proceeds to a step 540 in which the hitless migration sequence is provided to conventional ring control hardware. In a step 550, the ring control hardware causes the circuits to be rerouted in accordance with the hitless migration sequence. The method ends in an end step 560. Those skilled in the art should understand that, as circuits are added and dropped from the ring over time, the method 500 may be carried out again to perform further network engineering with respect to the ring.
Now, the performance of the algorithm will be explored for various objective functions. As pointed out earlier, no known heuristic exists to solve RRP. However, to provide a “loose” benchmark, the results from the algorithm will be contrasted with that of a known ring design algorithm that minimizes the size of the ring (ring sizing problem). This known ring design algorithm will be referred to hereinafter as “Ideal.”
Ideal yields routes within four times the optimal routing using techniques suggested in Cosares, et al. and Myung, et al., supra. The technique in Myung, et al., supra, will be used for ring loading to route circuits within a factor of two to the optimal. Then, Tucker's algorithm, as suggested in Cosares, et al., supra, will be used for slotting within another factor of two to the optimal.
However, Ideal is impractical for various reasons. First, Ideal does not produce a path migration sequence (and which may not exist at all). Second, Tucker's algorithm does not account for the alignment constraint. Thus, Ideal merely provides a loose theoretical bound and helps compare the ability of the algorithm and Ideal to minimize the ring size.
The results of the experiments are based on extensive simulations on a 16-node, STS-48 BLSR. Circuit requests were generated among any randomly chosen source-destination pair with the granularity also chosen randomly. Twenty percent of the requests were circuit deletions to mirror an operational network. The circuits were routed along the shortest path between source and destination and its slot was chosen in a first-fit manner (accounting for the alignment constraint). Requests were generated until a new request failed to be routed and this then formed the input to the algorithm and Ideal.
Two objective functions were used to evaluate the performance of the algorithm.
Experiment 1: Example Network Analysis
With reference to
Finally,
Experiment 2: Gains Analysis
While Experiment 1 demonstrated the mechanics of each heuristic on a single example network, it is worthwhile to quantify the gains in general.
Consider the SQM case, illustrated in
The LTM case, illustrated in
These results demonstrate the algorithm is effective and effectively engineers the network. For the SQM metric, it is able to create free spaces that balance both the length and granularity whereas the LTM is successful in clearing out long free spans. The LTM and SQM metrics require 21.76 and 17.84 circuit reroutes on the average respectively, which are within acceptable limits. Given that the gains brought about by these reroutes cause no service disruption, the algorithm is attractive.
Although the present invention has been described in detail, those skilled in the art should understand that they can make various changes, substitutions and alterations herein without departing from the spirit and scope of the invention in its broadest form.
Acharya, Swarup, Gupta, Bhawna, Srivastava, Anurag, Risbood, Pankaj
Patent | Priority | Assignee | Title |
Patent | Priority | Assignee | Title |
5608720, | Mar 09 1993 | PULSE COMMUNICATIONS, INC | Control system and operations system interface for a network element in an access system |
5745476, | Jul 16 1996 | AT&T Corp. | Errorless switching techniques in ring network |
5818842, | Jan 21 1994 | Alcatel Canada Inc | Transparent interconnector of LANs by an ATM network |
6188667, | Mar 29 1996 | Alcatel USA, Inc. | Transport interface for performing protection switching of telecommunications traffic |
6246668, | Jun 09 1998 | CIENA LUXEMBOURG S A R L ; Ciena Corporation | Hitless manual path switching using linked pointer processors |
6754175, | Apr 27 2000 | AT&T Corp.; AT&T Corp | Hitless method and apparatus for upgrading a network |
7130263, | Mar 31 2001 | Ericsson AB | Heterogeneous connections on a bi-directional line switched ring |
20020071392, | |||
20020118636, | |||
20020131368, | |||
20030048776, | |||
20030235153, |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Jul 14 2003 | ACHARYA, SWARUP | Lucent Technologies, INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 014861 | /0693 | |
Jul 14 2003 | GUPTA, BHAWNA | Lucent Technologies, INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 014861 | /0693 | |
Jul 14 2003 | SRIVASTAVA, ANURAG | Lucent Technologies, INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 014861 | /0693 | |
Aug 08 2003 | Alcatel-Lucent USA Inc. | (assignment on the face of the patent) | / | |||
Aug 13 2003 | RISBOOD, PANKAJ | Lucent Technologies, INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 014861 | /0693 | |
Nov 01 2008 | Lucent Technologies Inc | Alcatel-Lucent USA Inc | MERGER SEE DOCUMENT FOR DETAILS | 025410 | /0067 | |
Jul 22 2017 | Alcatel Lucent | WSOU Investments, LLC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 044000 | /0053 | |
Aug 22 2017 | WSOU Investments, LLC | OMEGA CREDIT OPPORTUNITIES MASTER FUND, LP | SECURITY INTEREST SEE DOCUMENT FOR DETAILS | 043966 | /0574 | |
May 16 2019 | WSOU Investments, LLC | BP FUNDING TRUST, SERIES SPL-VI | SECURITY INTEREST SEE DOCUMENT FOR DETAILS | 049235 | /0068 | |
May 16 2019 | OCO OPPORTUNITIES MASTER FUND, L P F K A OMEGA CREDIT OPPORTUNITIES MASTER FUND LP | WSOU Investments, LLC | RELEASE BY SECURED PARTY SEE DOCUMENT FOR DETAILS | 049246 | /0405 | |
May 28 2021 | TERRIER SSC, LLC | WSOU Investments, LLC | RELEASE BY SECURED PARTY SEE DOCUMENT FOR DETAILS | 056526 | /0093 | |
May 28 2021 | WSOU Investments, LLC | OT WSOU TERRIER HOLDINGS, LLC | SECURITY INTEREST SEE DOCUMENT FOR DETAILS | 056990 | /0081 |
Date | Maintenance Fee Events |
Jan 24 2011 | ASPN: Payor Number Assigned. |
Jun 27 2014 | M1551: Payment of Maintenance Fee, 4th Year, Large Entity. |
Aug 27 2018 | REM: Maintenance Fee Reminder Mailed. |
Jan 04 2019 | M1552: Payment of Maintenance Fee, 8th Year, Large Entity. |
Jan 04 2019 | M1555: 7.5 yr surcharge - late pmt w/in 6 mo, Large Entity. |
Aug 22 2022 | REM: Maintenance Fee Reminder Mailed. |
Dec 22 2022 | M1553: Payment of Maintenance Fee, 12th Year, Large Entity. |
Dec 22 2022 | M1556: 11.5 yr surcharge- late pmt w/in 6 mo, Large Entity. |
Date | Maintenance Schedule |
Jan 04 2014 | 4 years fee payment window open |
Jul 04 2014 | 6 months grace period start (w surcharge) |
Jan 04 2015 | patent expiry (for year 4) |
Jan 04 2017 | 2 years to revive unintentionally abandoned end. (for year 4) |
Jan 04 2018 | 8 years fee payment window open |
Jul 04 2018 | 6 months grace period start (w surcharge) |
Jan 04 2019 | patent expiry (for year 8) |
Jan 04 2021 | 2 years to revive unintentionally abandoned end. (for year 8) |
Jan 04 2022 | 12 years fee payment window open |
Jul 04 2022 | 6 months grace period start (w surcharge) |
Jan 04 2023 | patent expiry (for year 12) |
Jan 04 2025 | 2 years to revive unintentionally abandoned end. (for year 12) |